Whiskers in germanium transistors etc

[Paul Stenning]

I have been in email correspondence with a collector (Chris Walter) about the whiskers that develop in AF117 etc transistors. He initially contacted me offering a couple of photos showing these in an OC170 transistor he had opened up. His photos are attached and the main text of his accompanying email is here

Attached I send you two pictures from an OC170, which I found inside a Philips L3X09T from 1960.

It is very hard to take these pictures, since only a few of the many whiskers (I guess there are about 100 visible with naked eye) reflect the light at any given moment. The whole inside is covered with these whiskers whose lenghts (~2 mm long) and direction (around 90 degree to surface) do not vary very much. In sunshine and in artificial light I observed rainbow colouring of the hairs, which means their diameter must be in the order of the wavelength of visible light, that is 0.4 to 0.7 micrometer, which is about 1/200 th of a human hair. If I find time I will do some more observations, for example if the whiskers also grow into the grease.

No, I did not reassemble the transistor, since I broke two of the legs when opening it (I had to sacrifice another nice radio, to get an AF125 for replacement). But there is another reason: The transistor was intermittendly working before unsoldering it. I then applied the charged capacitor trick, not with 50 microF at 50 Volt, but with 200 microF and 15 Volt. I tied the three active legs together before. It turned out I had killed the base-emitter diode by doing the trick, this is when I decided to open it.

A while later he sent me a further email containing this website link http://nepp.nasa.gov/whisker/index.html. We are assuming that the transistor cases are made of tin or are tin plated internally.

Reading this website, in particular the Basic Info/FAQ section, shows that "tin whiskers" are a MUCH bigger problem than the irritating failure of transistors in a few old radios. It has caused an emergency recall of a batch of pacemakers, failure of sattelite control systems etc. It seems that the move to lead-free soldering etc could cause further problems as more tin is used in electronic assemblies.

[adibrook]

wow thats pretty fascinatign! I'v herd of NiCd batteries growing wiskers too. What actually causes it in transistors?

Thanks for sharing those photos!

[Paul Stenning]

Originally Posted by adibrook

What actually causes it in transistors?

It seems that nobody knows for sure - not even NASA with their huge research budgets.

[Darren-UK]

I have heard that during the 1960s impure germanium was used and thus it continues to 'grow' inside the transistor and somehow reacts with the metal(s) used for the screening can. For all I know, however, this could all be a load of codswallop.

Over the years I have kinda studied this problem and it does seem the whiskers are more likely to appear when a radio is left disused for long periods and much less so when in regular use.

[ukcol]

Originally Posted by darren-uk

Over the years I have kinda studied this problem and it does seem the whiskers are more likely to appear when a radio is left disused for long periods and much less so when in regular use.

That's a very interesting observation Darren and throws a new light on the process as far as I am concerned. I had always supposed (and not based on any experiments or observations) that the process of these whiskers growing may be an electrical one; that the whiskers may grow as the result of an electric field from a point of one electrical potential to another. Since you have observed that the process is more vigorous when the transistor is not in use this explanation is less likely. Has anyone observed whether the whisker grows from the case to the transistor, or the reverse, or if the whisker grows from both ends and meets in the middle?

[Paul Stenning]

The photos seem to suggest that the whiskers grow from the case rather than from the germanium.

[Darren-UK]

Hmmm.....that NASA site is very interesting and gives examples of whiskers growing from both tin and zinc, so I'm left wondering what other metals can also suffer this effect.

There is also an example of whiskers growing after 3 years of 'ambient storage' which does indeed imply, but not confirm, this is more likely to occur during disuse rather than use.

As PS pointed out, the whiskers do indeed appear to be eminating from the transistor case rather than the germanium. Question now is this; are these whiskers a property of, say, tin - or do they occur due to some sort of reaction between the tin and some other substance in very close proximity?

Must now find out if AF117 etc transistor cases are tin or zinc plated on the inside. If they are, why would they be ?

[krisgreen]

I am that Chris. The whiskers definitely grow from the case. As I said, there is rainbow colouring, changing colour even along one and the same fiber. That means the diameter is changing (narrowing down towards the end). Although measuring only 1/200th of a hair in diameter, the whiskers are still some tenthousand atomic layers across. There may be quite high electric fields near the tip, although the voltages used are low. I remember, however, to have read that such whiskers consist of a single monocrystal, and grow unidirectionally under the influence of the much higher internal electric field gradients alone.
I still have the opened transistor and am happy to follow suggestions what to do with it.

[Paul Stenning]

Originally Posted by darren-uk

Must now find out if AF117 etc transistor cases are tin or zinc plated on the inside.

I emailed the contact address on the website with brief details plus a copy of the email and photos from Chris Walter, and received the following reply:

Dear Paul,

We are pleased that our metal whisker www site has been useful to you and hopefully your colleagues as well once you've made them aware of it's location.

Would you and your colleagues be interested in having an "anecdote" of your observations posted within the NASA www site? I would simply use your note below as the foundation along with the 2 images you provided.

For examples of other whisker anecdotes we've posted, see:
http://nepp.nasa.gov/whisker/anecdote/index.html

Also, if any of the subject transistors could be provided to me, I could easily have a materials analysis performed to determine if these are tin, zinc, cadmium or other form of metal whiskers. If so, I could provide you with mailing information, etc. We'd gladly share any info we produce from such specimens with your club.

Sincerely,
Jay

P.S.-- I'm a little delayed in posting links (or anecdotes) I've heard from a few other "radio" enthusiasts related to their whisker experiences. Their interests are in 2-way (HAM) type radio applications in which they've noted whiskers (zinc and possibly tin) have impacted the performance of older models.

See for example -- http://www.repeater-builder.com/ge/mastrIIcasting.html
While the note on the above page identifies these as TIN whiskers, I have subsequently acquired info that leads me to believe this is misidentified and that they are really ZINC whiskers in this case.

Later I received a further email from another member of the team:

Dear Paul,

The two images you provided show a fine set of whiskers. I second Jay's remark that we could carry out an assay to determine the kind of metal; this is non-destructive to the part (but we might need to remove a whisker), and we would return the part to you along with a copy of the analysis.

The analysis instrument of choice is a Scanning Electron Microscope
(SEM) for imaging at powers up to ~10,000X, although mostly we would work at ~1,000X and even lower. This SEM has an EDAX attachment: the electrons in the scanned beam have sufficient energy that they drive inner-shell electrons out of the atoms at the surface of the object being examined. As other electrons flood back to fill the just-created vacancy, x-rays are emitted. The energy of these x-rays is uniquely related to the specific element, so we can read off the abundance of elements present on the surface of the specimen, with a spacial resolution of micrometers. (This does not make the specimen radioactive, or harm it.)

But the goo visible inside these transistor packages might outgas fast enough when placed into the SEM's vacuum chamber to interfere with the beam --- hard to tell from these images. If so, then we would extract a single whisker and examine it.

Best,
Henning Leidecker

I have several NOS AF115s here so I will take them up on their offer and send them a couple for analysis. I am not bothered whether they are returned or what state they are in.

I suspect the team at NASA are interested in getting hold of any samples they can to help with their research, which is why they are so keen to investigate this for us. Otherwise why would NASA be remotely interested in helping a group of vintage radio collectors?

I will also take them up on the offer of an "anecdote" on their website, if this is appropriate once we have further information on the material used.

[paulsherwin]

Originally Posted by Paul Stenning

I suspect the team at NASA are interested in getting hold of any samples they can to help with their research, which is why they are so keen to investigate this for us. Otherwise why would NASA be remotely interested in helping a group of vintage radio collectors?

Given the age of the shuttle fleet and the current emphasis on safety, I can well understand why resources are being directed at this problem. I imagine an accident caused by tin whiskers would be very difficult to explain subsequently.

Presumably the shuttle is full of 1970s metal cased transistors

Best regards, Paul

[Paul Stenning]

I have just replied to their email as below:

Dear Jay, Henning and team,

Thank you for your kind offer to analyse a sample to determine what type of whiskers we are experiencing. I have a number of these transistors that have never been used and have identified a few experiencing this problem. Could you please let me have a mailing address to send them to?

Because the whisker growth is internal I would assume the transistors will have to be opened somehow to examine the whiskers and analyse the internal surface of the case. I believe the case is soldered together but do not know whether the heat of desoldering would affect what we are trying to see.

In any event I can send you three or four to examine, and I do not need them to be returned. Therefore if you need to cut them open or otherwise damage them during examination, this is of no concern.

It would be interesting to know whether the growth also occurs in the area where the 'goo' (a type of grease?) is, or whether it is just in the areas containing air. Information from someone who has dismantled a couple of these implies that it is only in the areas of air. However it is likely that removing the grease would also remove the whiskers, so this may well be spurious.

We would also be interested in having an anecdote published on your website, however I think we should wait until after the analysis to see if this is appropriate, and to have more information to include.

Many thanks for your interest, and I hope you are able to gain a little useful information for your own research from this too.

Best regards,
Paul.

I have found four NOS AF114 transistors (from a batch of around 30) which have short circuits of a few tens of ohms between the case and either the collector or emitter, which I will send them. The rest of the batch are currently OK but of course it is only a matter of time.

Originally Posted by paulsherwin

I imagine an accident caused by tin whiskers would be very difficult to explain subsequently.

From this page http://nepp.nasa.gov/whisker/failures/index.htm and external pages it links to it is clear that a number of organisations have already had to explain why tin whiskers have caused the loss of satellites, nuclear reactor shutdowns, an emergency recall of a batch of pacemakers and recurring server faults in a data centre. As an example the loss of several Hughes satellites worth around $250 million each was exlained (in the media) along these lines:

Although the Solidaridad I failure is still being investigated, an earlier HS 601 processor defect could be the problem. The previous failure resulted from electrical shorts caused by internal tin-plated relay latching switches that act as on/off switches within the Spacecraft Control Processor. Hughes found that under certain conditions, a tiny, crystalline structure, less than the width of a hair, can grow and bridge a relay terminal to its case, causing an electrical short. Hughes is now using nickel-plated switches and improved processes, which appear to have corrected the processor defect on newer satellites. Engineers are currently studying data received from Solidaridad I to determine if the failure occurred in a similar fashion to previously recorded incidents.

[ukcol]

There's quite a bit of relevant information here:


http://www.calce.umd.edu/lead-free/t...ISKERRISKS.pdf

[GMB]

Over the years I have kinda studied this problem and it does seem the whiskers are more likely to appear when a radio is left disused for long periods and much less so when in regular use.

I'm surprised because I remember how some early computers suffered badly from this problem - and they were almost continuously powered.

FYI
Normal metals that you encounter in everyday life have very five crystal structures and are usually impure and also have been worked in some way so you may be surprised how different these metals behave when ultra-pure and single crystal. I was amazed the first time I handled a single crystal bar of pure tin...

At first it seems to have no strength at all - you expect a solid bar of metal but it is actually more like Plasticine - it cannot support its own weight and sort of flows and sags where unsupported. As it does this you see slip planes appearing as chunks of metal slide along the plane directions of the crystal. But this state doesn't last long because as it moves the crystal structure fines up and it quickly turns hard and more normal. You also notice as you handle it that it crackles - this sound is caused by a process called twinning in which single crystals suddenly switch their orientation. When the crystals are really big you can actually hear the atoms snapping into their new alignment.
It's dislocations in the crystal structure that prevents a material freely flowing and bending about. When in this pristine state the atoms can slip over each other so easily that you get this strange behaviour - and you can more readily appreciate how a fine whisker could be jacked out from a metal surface if some small cyclic force is applied to it.

[Hybrid tellies]

Just a further thought, mention has been made whether these whiskers grow whilst equipment is powered up or not.
I notice from Chris Walters pictures of the opened up OC171 transistors that the whiskers grow from the case inwards. Could this be down to a polarity thing as well as the type of metals used?
As far as I know the case (or screen) is connected to chassis which is positive in relation to the other electrodes particularly the collector which would explain why the most common failing of AF117 transistors is the collector screen shorts.
It would certainly be interesting to see what our friends at NASA make of those AF114's Paul has sent them.

[mikelect]

Originally Posted by GMB

It's dislocations in the crystal structure that prevents a material freely flowing and bending about. When in this pristine state the atoms can slip over each other so easily that you get this strange behaviour - and you can more readily appreciate how a fine whisker could be jacked out from a metal surface if some small cyclic force is applied to it.

This doesn't explain the same phenomena happening to gold which is an entirely different ductile metal

Mike

[Paul Stenning]

I have just received the following interesting email from Jay Brusse at Nasa:

Subject: The Verdict is.... they are TIN Whiskers inside AF114!!!

Dear Paul,

Sorry for our delay in analyzing the AF114 Germanium transistors you sent to us some weeks ago. I've attached some PRELIMINARY info (Power Point Slide Show) from our analysis of 1 of the 4 parts you sent to us. Our analysis to date confirms the following:

Metal whiskers are growing profusely from the internal surface of the TO package.

Materials analysis confirms that the whiskers are TIN.

The external surface of the can is also TIN (but no whiskers noted).

Due to the geometry of the opened package, we have not yet been able to document that the INTERNAL surface is tin, but it is a logical conclusion that this will be the case. Still we would need to dissect the can a bit further in order to allow our instrument detectors to "see" this surface. We are not yet pursuing this analysis as we first contemplate any other examinations we might like to do before more dissection.

Whiskers are bridging from case wall to the internal pins. Some nice dissection work by my co-worker, Mr. Chris Greenwell, has left shorting whiskers intact for photo-documentation.

Whiskers do grow readily through the 'grease' (suspect Silicone based?). We are contemplating techniques to more clearly photo-document this fact (photos included are not as good as we'd like to make this point) such as by cutting off the top of the already opened can to enable us to better illuminate the whiskers within the grease.

I hope you don't mind, but I am distributing this same initial information to a 'tin whisker' email forum I participate in. They have expressed a high degree of interest in "seeing" what I have "told" them thus far. For more permanent archival, I hope to post an "anecdote" of some form to the NASA whisker www site in the not too distant future. Our site is TEMPORARILY down for logistical maintenance matters (hope to be back soon?!?!)

Many thanks again, Paul, for your kind contribution. I certainly owe you one (or two or three depending on how much you like your liquor). I will keep you informed of any additional info we learn from your specimens.

Sincerely,

Jay


Jay Brusse
QSS Group at NASA Goddard
Sr. Components Engineer

The PowerPoint slideshow mentioned is too large to attach to this message (approx 2.5MB) so I have uploaded it to this new page on my Vintage Radio Information site. I have also created a PDF version for those who do not have PowerPoint installed, which is available on the same page.

Please take a look - some of the photos are fascinating!

[mikelect]

Thats great Paul very interesting please keep me posted of any more developments


Mike

[Sean Williams]

Crikey, these guys dont mess about do they!

Either that or they have far too much spare time, and desperatly need a hobby!

Very interesting though

Cheers
Sean

[Colin]

Interesting stuff ... and all the way from NASA too

I particularly like this bit ...

Due to the geometry of the opened package,

It's a great bit of NASA speak, I imagine the can was probably crushed by the hammer as they tries to open it

They probably used a laser

[Mike Phelan]

Truly amazing! It's a wonder these AF ever work!